lantern docs

lantern docs generates a complete, browsable documentation site from your agent project. It reads your code and configuration files statically — no LLM, no API keys.

Usage

# From the project root
cd my-agent-project
lantern docs

# From anywhere
lantern docs /path/to/my-agent-project

# Custom output directory
lantern docs /path/to/my-agent-project -o /path/to/output

# Generate and serve immediately
lantern web
lantern web /path/to/my-agent-project --port 9000

# Serve without regenerating
lantern web --no-generate

What Gets Generated

All files are written to <project>/docs/ by default.

File	Content
overview.md	Project snapshot, entrypoints, high-level flow
architecture.md	System map, key files, dependencies, environment variables
diagrams.md	Mermaid diagrams — agent-task graph, execution flow, sequence
agents.md	Agent roles, goals, backstories, tools, LLM, delegation flags
tasks.md	Task descriptions, assigned agents, expected outputs, context
runbook.md	Install, configure, run, static checks, troubleshooting
contact.md	Project contacts from pyproject.toml
index.html	Docsify site — publishable to GitHub Pages
_sidebar.md	Navigation sidebar
agentlantern-docs.html	Single-file HTML bundle (no server; Mermaid uses a CDN)

How It Works

1. Detection — AgentLantern scans pyproject.toml and the project structure to identify the framework
2. Analysis — the framework-specific analyzer parses agents, tasks, tools, LLM config, and env vars from code and YAML files (static AST parsing — no imports, no execution)
3. Rendering — a Markdown file is generated for each section, then assembled into a Docsify site

Sharing Documentation

Method	How
Local	lantern web → open http://localhost:9000
Single file	Send docs/agentlantern-docs.html — no server needed; diagrams require network access
GitHub Pages	Push docs/ and enable Pages in repository settings
Any web server	docs/ is a static site — works on Nginx, Vercel, Netlify

CI — Auto-generate on Push

# .github/workflows/docs.yml
name: Generate Docs

on: [push]

jobs:
  docs:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - uses: actions/setup-python@v5
        with:
          python-version: "3.11"
      - run: pip install "agentlantern @ git+https://github.com/brellsanwouo/agentlantern.git"
      - run: lantern docs
      - uses: actions/upload-pages-artifact@v3
        with:
          path: docs/

Quick Start (No API Keys Needed)

You can test lantern docs on a minimal CrewAI project without any LLM or API keys.

# 1. Create a scaffold project
pip install crewai
crewai create my_test_project
cd my_test_project

# 2. Generate documentation
pip install "agentlantern @ git+https://github.com/brellsanwouo/agentlantern.git"
lantern docs

# 3. Browse it
lantern web
# → http://localhost:9000

Supported Frameworks

Framework	Analysis level	What is extracted
CrewAI	Full	Agents, tasks, tools, LLM, delegation, memory, Mermaid diagrams
Google ADK	Full	Agents, tools, workflows, Mermaid pipeline diagrams
LangGraph	Full	Graphs, nodes, routes, tools and Mermaid diagrams
AutoGen	Full	Agents, teams, tools, runs and Mermaid diagrams
Smolagents	Full	Agents, managed agents, tools and Mermaid diagrams
OpenAI Agents SDK	Full	Agents, handoffs, tools, runners and Mermaid diagrams
Additional Python ecosystems	Dedicated	Agents/roles, tools, teams/workflows, factories and runs
Mastra / Vercel AI SDK / JS/TS	Dedicated	Agents, tools, workflows, npm scripts and diagrams
Semantic Kernel .NET/Java	Dedicated	Agent initializers, instructions and invocations
Other Python or JS/TS frameworks	Generic	Conventional agents, tools, relations and entrypoints